System and method of determining item storage strategy

1. A system for selecting items to be stored in a warehouse from a list of candidate items, the system comprising a computing device, the computing device comprising a processor and a storage device storing computer executable code, wherein the computer executable code, when executed at the processor, is configured to: assign scores to the respective candidate items based on an item selection model; and select items from the candidate items to be stored in the warehouse according to the assigned scores under a capacity limit of the warehouse, wherein the item selection model is configured with an Artificial Neural Network (ANN), having item data for an item as input and a score for that item as output, the score representing a probability that the item is to be stored in the warehouse, and wherein the ANN is trained by historical orders, each of which includes one or more items, under a constraint of the capacity limit of the warehouse with an objective of maximizing a probability that the orders have their respective item(s) all stored in the same warehouse.

2. The system of claim 1 , wherein the ANN is configured with two or more hidden layers.

3. The system of claim 1 , wherein the computer executable code, when executed at the processor, is further configured to represent the item data for the item as a vector corresponding to the item.

4. The system of claim 3 , wherein the vector for the item comprises element(s) associated with a unique identifier (ID) of the item and element(s) associated with a property of the item.

5. The system of claim 4 , wherein the vector for the item further comprises an element associated with sales data of the item and/or an element associated with promotion data of the item.

6. The system of claim 4 , wherein the property comprises at least one of category or brand of the item.

7. The system of claim 4 , wherein the computer executable code, when executed at the processor, is further configured to update mapping from the ID to its corresponding element(s) and mapping from the property to its corresponding element(s) based on the training.

8. The system of claim 4 , wherein the element(s) associated with the ID or the element(s) associated with the property is/are randomized at initialization of the ANN.

9. The system of claim 1 , wherein a probability that an item is to be stored in the warehouse is a sigmoid function of the score assigned to that item, that is, the probability p item and the score θ item satisfy: p item = sigmoid ⁡ ( θ item ) = 1 1 + e - θ item .

10. The system of claim 1 , wherein a probability that an order has its items all stored in the same warehouse is a production of probabilities that the respective items included in this order are to be stored in the warehouse, and the ANN is configured to maximize a sum of the probabilities of the respective orders.

11. A method of selecting items to be stored in a warehouse from a list of candidate items, the method comprising: establishing an item selection model with an Artificial Neural Network (ANN), having item data for an item as input and a score for that item as output, the score representing a probability that the item is to be stored in the warehouse; training the item selection model with historical orders, each of which includes one or more items, under a constraint of a capacity limit of the warehouse with an objective of maximizing a probability that the orders have their respective item(s) all stored in the same warehouse; assigning scores to the respective candidate items based on the trained item selection model; and selecting items from the candidate items to be stored in the warehouse according to the assigned scores under the capacity limit of the warehouse.

12. The method of claim 11 , wherein the ANN has two or more hidden layers.

13. The method of claim 11 , further comprising representing the item data for the item as a vector corresponding to the item.

14. The method of claim 13 , wherein the vector for the item comprises element(s) associated with a unique identifier (ID) of the item and element(s) associated with a property of the item.

15. The method of claim 14 , wherein the vector for the item further comprises an element associated with sales data of the item and/or an element associated with promotion data of the item.

16. The method of claim 14 , wherein the property comprises at least one of category or brand of the item.

17. The method of claim 14 , further comprising updating mapping from the ID to its corresponding element(s) and mapping from the property to its corresponding element(s) based on the training.

18. The method of claim 14 , wherein further comprising randomizing the element(s) associated with the ID or the element(s) associated with the property at initialization of the ANN.

19. The method of claim 11 , wherein a probability that an order has its items all stored in the same warehouse is a production of probabilities that the respective items included in this order are to be stored in the warehouse, and the training is performed by maximizing a sum of the probabilities of the respective orders.

20. A non-transitory computer readable medium storing computer executable code, wherein the computer executable code, when executed at a processor, is configured to: establish an item selection model with an Artificial Neural Network (ANN), having item data for an item as input and a score for that item as output, the score representing a probability that the item is to be stored in the warehouse; train the item selection model with historical orders, each of which includes one or more items, under a constraint of a capacity limit of the warehouse with an objective of maximizing a probability that the orders have their respective item(s) all stored in the same warehouse; assign scores to the respective candidate items based on the trained item selection model; and select items from the candidate items to be stored in the warehouse according to the assigned scores under the capacity limit of the warehouse.